When Actinium Stops Being the Bottleneck

The Paradox

There are more Ac-225 production projects in the US and Europe than at any point in the isotope's history, and there are still no commercially scheduled Ac-225 treatments, because no Ac-225 therapeutic is approved in either market. Both statements are true at 31 July 2026, and holding them together is the whole of the argument.

The public conversation treats isotope scarcity as the problem to be solved. Solve supply, the reasoning goes, and alpha therapy scales. We have mapped that supply picture before, including who actually controls the alpha emitter bottleneck. The industrial evidence points somewhere less convenient. Bulk isotope can stop being the binding constraint while the system remains just as hard to run, because the things that actually convert a curie into a treated patient sit downstream of the accelerator and expand on much slower clocks.

The shortage can end before the system scales.

The Industry Is Measuring the Wrong Thing

Capacity in this sector arrives in stages, and almost all public reporting stops at the first one. A useful ladder runs: announced, commissioned, demonstrated, operating, GMP or otherwise regulatory usable, reliably available and commercially allocated, product-specific finished dose, administered treatment. Every step loses projects, volume or availability. Only the last two determine whether anyone can be scheduled. It is the same gap between a press release and a bookable slot that produces ghost capacity in contract manufacturing.

The evidence illustrates each rung. TerraPower Isotopes and PanTera have broken ground on facilities targeting the end of the decade, which is a financing and commitment signal and nothing more. Nusano reported first beam on target at partial energy in June 2026, which is commissioning. Ionetix announced a first commercial-scale batch in June 2024, roughly nineteen months after its NRC materials licence, which is demonstration rather than a multi-quarter cadence. NorthStar states routine weekly radiochemical-grade output from Q4 2025 and an accepted Type II drug master file, which crosses several gates at once but is still company disclosure rather than published released activity. The Department of Energy reports more than 1,102 mCi distributed cumulatively, with typical batches in the 50 to 100 mCi range and larger batches described as a forward plan.

None of those numbers can be responsibly converted into patient counts. One unit of bulk activity passes through sponsor allocation, decay, labeling yield, trace-metal competition, QC sampling, batch rejection, shipping, calibration, per-administration activity and the number of cycles a patient actually completes. Those factors are product-specific and route-specific, and most of them are undisclosed. A curie-to-patient conversion built on public data is arithmetic performed on the wrong inputs.

So the practical unit is different: on-time released doses matched to staffed, alpha-ready slot-hours. It is awkward, and that is its value. It forces decay, allocation, labeling yield, QC rejection, delivery, licensing, staffing, imaging, payer authorisation and patient readiness into one frame instead of eight separate dashboards.

The Clinical Clock Has Moved

The demand side is later and narrower than the isotope-shortage narrative implies. At the cutoff, four public pivotal trials carry the commercially consequential weight.

ACTION-1 is a randomised Phase Ib/III study of RYZ101 in SSTR-positive gastroenteropancreatic neuroendocrine tumours that progressed after Lu-177 somatostatin-receptor therapy, with 338 estimated participants. Bristol Myers Squibb's latest formal guidance places registrational data in 2027, superseding older 2026 guidance. Novartis has two AAA817 programmes: PSMAcTION, a Phase II/III in 443 participants with a registry primary completion in June 2028, and AcTFirst, a Phase III in earlier-line mCRPC with 940 participants, up to six 10 MBq cycles and a registry primary completion in September 2028. AstraZeneca's VECTRA-01 is a Phase III in PSMA-positive mCRPC with 670 participants and a primary completion in February 2029.

Read together, the first credible approval cluster falls in 2028 to 2030, conditional on positive data, CMC readiness and prompt review. That is an analyst inference, not a schedule.

There is a second and less comfortable point. Programme counts overstate independent probability. AAA817, AZD2265 and Bayer's Ac-PSMA agent share target expression, PSMA-PET based selection, advanced-prostate referral pathways and the same class risks in salivary, marrow and renal tissue. RYZ101 and the Pb-212 SSTR programmes share somatostatin biology, prior-PRRT populations and a limited disease pool. Two PSMA successes are one demand branch, not two. A genuine clinical breakout needs at least two biologically and commercially distinct clusters, and that raises rather than lowers the risk of overbuilding upstream.

Finished-Dose Manufacturing Is the Hidden Hinge

Bulk isotope does not treat anyone. Between a released curie and an administered dose sits a set of steps that can fail entirely independently of isotope production: source qualification and comparability, product-specific conjugation and formulation, low-activity assay, sterility assurance, route-specific impurity methods, batch release and decay-aware distribution. This is the part of the chain we have called execution under decay, and it is where most radiopharma manufacturing plans quietly come apart.

IAEA guidance sets out why this is harder than the beta-emitter equivalent. Dedicated enclosed handling, contamination control, alpha-sensitive measurement and route-specific impurity and waste characterisation are all material requirements. There is no stable actinium isotope to anchor analytical validation. Daughter ingrowth and equilibration add time. Trace iron, copper or zinc can depress DOTA labeling yield or lose a batch outright.

The warning case comes from the neighbouring isotope. ARTBIO voluntarily paused its AB001 ARTISAN trial in April 2026 after process-related variability. That is a Pb-212 programme and the failure mechanism cannot be transferred mechanically to Ac-225 products. What it demonstrates is narrower and still uncomfortable: a nominally available alpha isotope does not rescue an unstable product process. Availability upstream and control downstream are different problems.

Under a single narrow launch, this hinge stays regional and product-specific. Under a breakout, several products compete for the same qualified chemists, QC instruments, release staff and logistics windows, and a physically complete facility can sit idle while methods, process performance qualification or sponsor comparability remain unresolved.

Lu-177 Is an Option on Alpha Capacity

The Lu-177 franchises have built something genuinely valuable: licensed nuclear-medicine departments, established target-imaging pathways, multidisciplinary referral relationships, payer familiarity and working radiopharmacy processes. Novartis reported more than 880 US Pluvicto sites and more than 650 ex-US sites in July 2026, with Q2 Pluvicto sales of USD 651 million, up 43% at constant currencies. An April 2026 access page reported 851 US RLT treatment sites against 129 in 2018.

Those are company-reported figures, and they establish that a manufacturer can recruit nodes and grow a franchise. They do not disclose administrations per site, spare slots, cycle completion or any alpha capability at all. Proximity is not appointment availability, which is the same distinction that makes the clinical delivery network the phantom market underneath every radiopharma forecast.

The additional requirements for alpha are specific. Ac-specific authorisation language, sometimes through state or national review. Stronger internal-contamination discipline and alpha-sensitive probes. MBq-scale activity measurement with daughter equilibrium handled inside a validated method, where the mature Lu workflow is calibrated at GBq scale. Route-dependent waste streams, since Ac-227 content and daughter behaviour differ by production route. Image-based dosimetry is harder because the imaging signal is limited. And a senior alpha lead with supervised cases behind them, which is the least capital-elastic item on the list and the reason a Head of Radiochemistry search rarely resolves quickly.

A February 2025 NRC event report describes an Ac-225 FPI-2265 administration in which a pump leak resulted in roughly 80 of a prescribed 289 microcuries being delivered. One event is not a frequency estimate. It is a reminder that low activity does not mean low operational consequence.

A Lu-177 centre is an option on alpha capacity, not alpha capacity itself.

Whether the option gets exercised depends on licence, assay, waste, training and slot availability, not on the site count.

The Lead-Time Mismatch

The planning ranges make the shape of the problem obvious. These are model assumptions used to compare capabilities against each other, not measured industry averages, and they should be challenged as such.

Brownfield Ac production runs 2 to 4 years, greenfield 3 to 6. Product-specific finished dose at an existing site runs 1 to 3 years, a new GMP site 3 to 5. A Lu-experienced centre reaching genuine alpha readiness runs 0.5 to 1.5 years, a new RLT centre 1 to 3. Licence amendments and procedures span 0.25 to 2 years depending on jurisdiction and whether the programme already exists. Reimbursement access runs 0.5 to 3 years across the US and Europe. Referral networks take 1 to 5 years of repeated institutional behaviour. Cross-training an experienced team takes 0.5 to 2 years. Forming genuinely new specialists takes 4 to 8 years or longer.

The workforce evidence sits underneath those last two lines. US BLS projects 3% growth in nuclear medicine technologist employment from 2024 to 2034, around 900 openings a year across all nuclear medicine. ASRT reported a 12.6% NMT vacancy rate in 2025 against an average of 4.1 budgeted FTEs per responding department. AAPM stated that nearly one-third of medical physicists would reach typical retirement age within five years. An EANM survey available in February 2026 found open positions in 80.2% of responding departments, though from 239 respondents at an estimated response rate near 3%, so directional rather than a population estimate. Those numbers are the arithmetic behind the three roles that are becoming near-impossible to fill.

The strategic trap follows directly. Isotope plants can be financed before demand exists, and therefore risk arriving too early if the correlated programmes fail. Human and institutional capacity is difficult to build before demand is visible, and therefore risks arriving too late if several programmes succeed. It is the clearest illustration we have of why, in complex modalities, capital is not the constraint.

Four Possible Futures

Pipeline disappointment. Most programmes fail or stay narrow. Ac-225 stops being scarce, but because demand never materialised. The constraint sits in clinical evidence and bankable demand through around 2030, then in utilisation and unit economics. Target-specific plants and suites are the stranded assets.

Managed expansion. One non-PSMA and one PSMA programme succeed, mostly in filtered post-Lu populations, and at least two supply chains qualify. Supply constraint gives way to released supply and then finished dose around the turn of the decade, with centres, workforce, referral and national access dominating through the early 2030s.

Clinical breakout. An earlier-line programme lands alongside a biologically independent target, and tolerability supports repeated dosing. Usable Ac and CMC bind through 2028, finished dose in 2029, centres and workforce from 2030, and workforce, imaging, referral and access thereafter. Bulk isotope may catch up in nameplate before finished dose becomes reliable.

Upstream disappointment. Clinical data are positive but precursor access, target recovery, yields, commissioning or qualification slip. Usable Ac remains the binding layer through the early 2030s, and downstream investment sits underused because it cannot be fed.

The underlying analysis does not attach probabilities to these, and neither will I. Public evidence does not support them.

What Would Prove the Model Right or Wrong

The monitoring is operational, not promotional. Released lots and on-time fill. Source qualification and referenced drug master files. Batch rejection rate and release time. Active centres rather than listed centres, measured as the share treating in the last 90 days. Administrations and completed cycles against planned cycles. Cancellation causes separated into isotope, batch release, shipping, payer, staff, slot, laboratory and patient reasons. Workforce vacancy, time-to-fill and supervised alpha cases. Referral-to-consult interval and imaging-to-dose conversion. Payer authorisation time, denial rate and net collection.

The single most useful of these is the cancellation reason code. When isotope and release deferrals fall for several consecutive quarters while staff, slot, imaging, licence and payer delays rise, the system has moved from upstream scarcity to downstream absorptive capacity. No supplier announcement will tell you that.

One qualification carries the whole structure. All of this infrastructure is a bet that the clinical pipeline produces treatments with meaningful, durable benefit and tolerable repeat dosing. If the pipeline disappoints, Ac-225 will cease to be scarce because demand never developed. If it works, the isotope may well become available before the system is ready to use it.

Where This Lands for Operators

Every layer in the explorer above that sits downstream of the accelerator resolves into a hiring decision at some point. Finished-dose release is a qualified-person and QC problem. Alpha readiness at a treatment centre is a supervised-cases problem. Referral conversion is an institutional-relationship problem owned by people who have done it before. None of those can be bought forward at the speed a hot cell can.

This is where our work sits. ProGen Search runs retained searches across radiopharma for exactly these seats: radiochemistry and CMC leadership, quality and qualified-person capability, technical operations, and the operating leadership that has to run a business on a decay clock. If you are financing capacity against a 2029 or 2030 launch, the question worth asking early is not whether the isotope will be there. It is whether the people who convert it into a treated patient will be.

This article is independent market intelligence and not investment, clinical or regulatory advice. It draws on public registry entries, regulator records, company disclosures and professional-body surveys believed reliable at an evidence cutoff of 31 July 2026, and covers the United States and Europe. Company-reported figures are company-reported, not verified operating output. The scenario framework and the lead-time ranges are planning constructs for comparing capabilities against each other, not forecasts, and carry no probabilities. Company names and trademarks are the property of their respective owners. © 2026 ProGen Search Limited.